Electronic medical record pre-generation method and device
Through the combination of ASR model and medical big model, real-time audio translation of doctor-patient dialogue and electronic medical record generation is achieved, solving the problem of low efficiency in generation of traditional electronic medical record, improving the efficiency and accuracy of medical data processing, and reducing the work pressure of doctors and Patient waiting time.
Patent Information
- Application Number
- CN202411002564.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-07-25
- Publication Date
- 2024-11-22
AI Technical Summary
The traditional electronic medical record generation method relies on manual input by doctors, which is inefficient and error-prone, and cannot effectively understand and handle complex doctor-patient conversations, causing doctors to spend a lot of time on medical record entry and editing, reducing work efficiency and extending patients' waiting time. .
By training the ASR model for real-time audio translation and tuning of doctor-patient conversations, electronic medical records are generated based on the medical big model in fine-grained manner, multi-source asynchronous management model is used to organize patient personal electronic medical records, and highly available medical records are generated through optimization strategies to reduce doctor reviews Work.
Solve the problem of cumbersome and repetitive work in writing electronic medical records, improve work efficiency, shorten the waiting time of patients, improve the accuracy and standardization of electronic medical records, alleviate the problems of medical disputes and medical data management, and improve the hospital. Overall medical efficiency.
Smart Images

Figure CN119007902A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of auxiliary medical technology, and specifically provides an electronic medical record pre-generation method and device. Background Art
[0002] With the continuous development of 5G technology and the continuous maturity of AI technology, the combination of 5G and AI is one of the important keys to accelerate technological development and realize new digital services. Among them, in the medical process, doctor-patient communication is an indispensable part of the medical process. The value of doctor-patient dialogue far exceeds that of dialogue between people in other fields. How to better use doctor-patient dialogue and extract and collect information from doctor-patient dialogue has become a key link in promoting medical efficiency.
[0003] At the same time, the collection and analysis of doctor-patient dialogues is also conducive to improving doctors' communication skills. Good doctor-patient communication not only helps to better diagnose and treat patients, but also plays an important role in reducing medical disputes. In recent years, the development of electronic medical records is an important part.
[0004] However, the traditional way of generating electronic medical records mainly relies on manual input from doctors, which is inefficient and prone to errors. Although some automated electronic medical record generation systems have appeared on the market, these systems mostly rely on rules and templates and cannot fully understand and process complex medical conversations, causing doctors to spend a lot of time on medical record entry and editing, reducing work efficiency and prolonging patients' waiting time. In doctor-patient communication, the conversation between doctors and patients contains a wealth of medical information, which is crucial for generating electronic medical records.
[0005] Therefore, how to effectively extract useful medical information from doctor-patient conversations and convert it into structured electronic medical record data has become an urgent problem to be solved. Summary of the invention
[0006] The present invention aims at solving the above-mentioned deficiencies of the prior art and provides a highly practical method for pre-generating electronic medical records.
[0007] A further technical task of the present invention is to provide an electronic medical record pre-generation device that is reasonably designed, safe and applicable.
[0008] The technical solution adopted by the present invention to solve its technical problem is:
[0009] A method for pre-generating electronic medical records. First, based on real-time doctor-patient conversations and offline doctor-patient conversation audios, an ASR model is trained to provide real-time translation and tuning capabilities for doctor-patient conversation audios, and also to provide offline audio translation management and medical record self-generation.
[0010] Then, the medical records are divided into the time and place of the main complaint and current medical history according to the medical industry standards, the main symptoms and their development and changes, the accompanying symptoms related to the main symptoms, and the detailed process and results of the examinations and treatment measures received by the patient in and outside the hospital before coming to the hospital for treatment;
[0011] Finally, a multi-source asynchronous management model for organizing patients' personal electronic medical records was adopted to solve patients' medical problems.
[0012] Furthermore, after the audio is translated, the corpus is initially optimized based on the medical big model. At the same time, through in-depth analysis of the medical record rules of a single department, the contents of the electronic medical record are generated in a fine-grained manner based on the doctor-patient dialogue corpus. At the same time, the medical record generation results are scored based on the electronic medical record evaluation template.
[0013] Furthermore, in general scenarios, through the optimization strategy of selecting the best of three, the electronic medical records are generated three times, the one with the highest score is obtained, and finally a highly usable electronic medical record of the patient and other relevant patient information is pre-generated;
[0014] After the electronic medical record is generated, the doctor needs to review it and make minor adjustments or no adjustments are required.
[0015] Furthermore, the detailed process and effects of the examinations and treatment measures received in and outside the hospital, including the patient's mental state, sleep, diet, urination, defecation, and weight changes since the onset of the disease;
[0016] Past history, personal history, history of exposure to toxic substances, dust, radioactive materials, information on prostitution history, marital history, menstrual and reproductive history, family history, whether there are hereditary diseases or other diseases that may be related to the current disease in the family, physical examination, auxiliary examinations and preliminary diagnosis.
[0017] Furthermore, the medical records are finely divided and detailed analysis is performed on each point in the medical records. The medical record content is accurately generated based on the doctor-patient dialogue through a large model. The doctor only needs to review the generated results to complete the patient's medical record writing.
[0018] Furthermore, the multi-source asynchronous management of the patient's personal electronic medical record organization mode establishes a data source for each department and examination room within the hospital, and asynchronously monitors the real-time diagnosis results of multiple data sources within the hospital through the patient's consultation ID.
[0019] Furthermore, after obtaining the diagnosis result, an asynchronous task is started and synchronized to the patient's electronic medical record, so that the consulting doctor can know the patient's examination results in the first time and the patient's medical problem can be solved.
[0020] An electronic medical record pre-generating device comprises: at least one memory and at least one processor;
[0021] The at least one memory is used to store a machine-readable program;
[0022] The at least one processor is used to call the machine-readable program to execute a method for pre-generating an electronic medical record.
[0023] Compared with the prior art, the electronic medical record pre-generation method and device of the present invention has the following outstanding beneficial effects:
[0024] The present invention solves the problem of tedious and repetitive work in writing electronic medical records, relieves doctors' work pressure, improves work efficiency, shortens patients' waiting time, and prevents medical disputes. At the same time, it improves the accuracy and standardization of electronic medical records, solves the problem of medical data governance, coordinates in-hospital data, links multi-source data, improves data utilization, alleviates the problem of mismatch between medical and patient resources, establishes a medical assembly line based on the patient's medical process, and improves the overall medical efficiency of the hospital. BRIEF DESCRIPTION OF THE DRAWINGS
[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.
[0026] Attached Figure 1 It is a schematic diagram of a framework of an electronic medical record pre-generation method;
[0027] Attached Figure 2 It is a flowchart of an electronic medical record pre-generation method. DETAILED DESCRIPTION
[0028] In order to enable those skilled in the art to better understand the solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0029] A best embodiment is given below:
[0030] like Figure 1-2 As shown, a method for pre-generating an electronic medical record in this embodiment first provides real-time translation and tuning capabilities of the audio of the doctor-patient conversation based on the real-time doctor-patient conversation and the offline doctor-patient conversation audio by training the ASR model, and also provides translation management of the offline audio and self-generation of medical records;
[0031] Then, the medical records are divided into the time and place of the main complaint and current medical history according to the medical industry standards, the main symptoms and their development and changes, the accompanying symptoms related to the main symptoms, and the detailed process and results of the examinations and treatment measures received by the patient in and outside the hospital before coming to the hospital for treatment;
[0032] Finally, a multi-source asynchronous management model for organizing patients' personal electronic medical records was adopted to solve patients' medical problems.
[0033] Among them, based on real-time doctor-patient conversations and offline doctor-patient conversation audio, the ASR model trained with 60,000 hours of corpus provides real-time translation and tuning capabilities for doctor-patient conversation audio, while also providing offline audio translation management and self-generation of medical records.
[0034] After audio translation, the corpus is initially tuned based on the medical big model to reduce the impact of irrelevant information on the generation of electronic medical records and improve the efficiency of medical information in doctor-patient dialogues. At the same time, through the in-depth analysis of the medical record rules of a single department, the contents of the electronic medical record are generated in a fine-grained manner based on the doctor-patient dialogue corpus, and the medical record generation results are scored based on the electronic medical record evaluation template. In general scenarios, through the optimization strategy of selecting the best of three, three electronic medical records are generated to obtain the one with the highest score, and finally highly usable patient electronic medical records and other related patient information are pre-generated.
[0035] After the electronic medical record is generated, the doctor only needs to make minor adjustments or no adjustments after review, which saves more than 80% of the work tasks for doctors in the electronic medical record writing process and effectively solves the problem of tedious and repetitive work in electronic medical record writing. The following is a timing diagram of online and offline generation of electronic medical records.
[0036] According to the standards of the medical industry, the medical records are divided into the main complaint (the main symptoms and signs of this visit and their duration), the current medical history, that is, the time and place of the patient's onset (only the main complaint), the main symptoms and their development and changes since the onset of the disease, the accompanying symptoms related to the main symptoms, and the detailed process and effects of the patient's examinations and treatment measures in and outside the hospital before coming to the hospital. The patient's mental state, sleep, diet, urination, and weight changes since the onset of the disease.
[0037] There is also past medical history (whether the patient has a history of chronic diseases, infectious diseases, vaccinations, surgery and trauma, blood transfusions, drugs and food allergies), personal history, history of exposure to toxic substances, dust, radioactive materials, information on travel history, marital history, menstrual and reproductive history, family history of disease conditions, health status, cause of death (if applicable, and whether there are hereditary diseases or other diseases in the family that may be related to the current disease), physical examination (the patient's systemic and local physical examination results), auxiliary examinations (the results of laboratory tests, imaging examinations and other special examinations completed before the patient was admitted to the hospital), and preliminary diagnosis (the patient's preliminary diagnosis conclusion).
[0038] We split the medical records into fine-grained parts, conduct detailed analysis on each point in the medical records, and use a large model to accurately generate medical record content based on doctor-patient dialogue. We conservatively estimate that we will eventually generate a usable medical record with a score of more than 80. Doctors only need to review the generated results to complete the patient's medical record writing. The above design can greatly relieve doctors' work pressure, improve work efficiency, shorten patients' waiting time, prevent medical disputes, and improve the accuracy and standardization of electronic medical records.
[0039] The multi-source asynchronous management mode for personal electronic medical records of patients establishes a data source for each department and examination room within the hospital, asynchronously monitors the real-time diagnosis results of multiple data sources within the hospital through the patient's consultation ID, and after obtaining the results, starts an asynchronous task and synchronizes them to the patient's electronic medical record. This allows the consulting doctor to understand the patient's examination results at the first time, so that the patient's medical problems can be solved faster and better.
[0040] Based on the above method, an electronic medical record pre-generation device in this embodiment includes: at least one memory and at least one processor;
[0041] The at least one memory is used to store a machine-readable program;
[0042] The at least one processor is used to call the machine-readable program to execute a method for pre-generating an electronic medical record.
[0043] The above-mentioned specific implementations are only specific cases of the present invention. The patent protection scope of the present invention includes but is not limited to the above-mentioned specific implementations. Any technical solutions that comply with the claims of the present invention and any appropriate changes or substitutions made by ordinary technicians in the relevant technical field shall fall within the patent protection scope of the present invention.
[0044] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for pre-generating electronic medical records, It is characterized in that First, based on real-time doctor-patient conversations and offline doctor-patient conversation audio, the ASR model is trained to provide real-time translation and tuning capabilities for doctor-patient conversation audio, while also providing offline audio translation management and self-generation of medical records. Then, the medical records are divided into the time and place of the main complaint and current medical history according to the medical industry standards, the main symptoms and their development and changes, the accompanying symptoms related to the main symptoms, and the detailed process and results of the examinations and treatment measures received by the patient in and outside the hospital before coming to the hospital for treatment; Finally, a multi-source asynchronous management model for organizing patients' personal electronic medical records was adopted to solve patients' medical problems.
2. The method for pre-generating an electronic medical record according to claim 1, It is characterized in that After the audio is translated, the corpus is initially optimized based on the medical big model. At the same time, through in-depth analysis of the medical record rules of a single department, the contents of the electronic medical record are generated in a fine-grained manner based on the doctor-patient dialogue corpus. At the same time, the medical record generation results are scored based on the electronic medical record evaluation template.
3. The method for pre-generating an electronic medical record according to claim 2, It is characterized in that In general scenarios, the best-of-three optimization strategy is used to generate electronic medical records three times, obtain the one with the highest score, and finally pre-generate highly usable patient electronic medical records and other relevant patient information; After the electronic medical record is generated, the doctor needs to review it and make minor adjustments or no adjustments are required.
4. The method for pre-generating an electronic medical record according to claim 3, It is characterized in that Detailed information on the examinations and treatments received in and outside the hospital and their effects, including the patient's mental state, sleep, diet, bowel movements, and weight changes since the onset of the disease; Past history, personal history, history of exposure to toxic substances, dust, radioactive materials, information on prostitution history, marital history, menstrual and reproductive history, family history, whether there are hereditary diseases or other diseases that may be related to the current disease in the family, physical examination, auxiliary examinations and preliminary diagnosis.
5. The method for pre-generating an electronic medical record according to claim 4, It is characterized in that Medical records are split into fine-grained parts, and detailed analysis is performed on each point in the medical records. The medical record content is accurately generated based on the doctor-patient dialogue through a large model. Doctors only need to review the generated results to complete the patient's medical record writing.
6. The method for pre-generating an electronic medical record according to claim 5, It is characterized in that The multi-source asynchronous management mode of organizing the patient's personal electronic medical records establishes a data source for each department and examination room within the hospital, and asynchronously monitors the real-time diagnosis results of multiple data sources within the hospital through the patient's consultation ID.
7. The method for pre-generating an electronic medical record according to claim 6, It is characterized in that After obtaining the diagnosis results, an asynchronous task is started and synchronized to the patient's electronic medical record, so that the consulting doctor can know the patient's examination results at the first time and the patient's medical problem can be solved.
8. An electronic medical record pre-generation device, It is characterized in that include: at least one memory and at least one processor; The at least one memory is used to store a machine-readable program; The at least one processor is configured to call the machine-readable program to execute the method according to any one of claims 1 to 7.
